Tobacco Documents Online

PHILIP MORRIS U. S. A. INTER-OFFICE CORRESPONDENCE RICHMOND, VIRGINIA it•JE-C E tY1rD APR i 1982 $. K NELSON To: . Mr. H. G. Daniel Date: March 25, 1982 From: . M. E. Johnston Subject: . Still More on the Price Elasticity of Cigarettes In two previous memos on this subject,l I summarized two papers dealing with the price elasticity of cigarettes, the conclusions of which were that price has its greatest effect on the young, particularly young males, and that price acts through its effect on reducing the proportion who smoke to a much greater extent than in reducing the cigarette consumption of committed smokers. The authors of these two papers concluded that the price elasticity of cigarettes was -0.42, i.e., that a ten percent increase in price would cause a 4.2 percent decline in sales. This is within the range (-0.1 to -1.5) found by most investigators. In this memo I report on the results of a state-by-state analysis of the effect of price increases on cigarette sales and conclude that price elasticities have been declining over time and are probably now about -0.2. I have continued to pursue the subject of the price elasticity of cigarettes for a variety of reasons: (1) I am convinced that the issue of an increase in the federal excise tax on cigarettes is not dead. There is still considerable support for it in Congress, in spite of the President's disavowal of it. (2) States will be feeling an increasing financial pinch as a result of the "New Federalism," and will undoubtedly raise excise 'taxes. Indeed, this has already begun. After a rash of increases between 1955 and 1971 there came a period of very few increases (Chart 1), but in 1981 six states raised their excise taxes, the greatest number since 1971. In addition, about 21 other states have tax bills pending.2 (3) I discovered that, in general, the newer studies of cigarette price elasticities show elasticities that are lower than the older studies. (4) Most important, people keep saying to me, in effect: "You economists keep talking about price elasticities, but sales continue to increase even though the prices also increase. What gives ?" 1. "Teenage Smoking and the Federal Excise Tax on Cigarettes," September 17, 1981, and "Cigarette Price Elasticities and the Implications for Philip Morris," January 5, 1982. 2. The reason for the consistent peaks in odd years is that many state legislatures meet only in odd years, and others have longer sessions in odd than in even years, while but one legislature (Kentucky) meets only in even years. This pattern should moderate in the future, as now only 14 state legislatures meet biennially. In the 1960's, 30 did so.

2 Just What is Price Elasticity? A couple of definitions are in order. (1) Price elasticity is the change in the demand for a product that results from a change in the price of that product. It is almost always negative, indicating that increasing prices reduces demand. An elasticity of 0.0 would indicate perfect inelasticity, i.e. that an increase in price would have no effect on demand, while an elasticity of -1.0 would indicate that a ten percent increase in price would reduce demand by ten percent. (2) Cross elasticity is the change in the sales of one product that results from a change in the price of some other product, e.g. when the price of beef goes up, people buy less beef so sales of chicken go up. Price elasticity works in two ways: (1) A change in price can cause a change in the number of people who use the product, or (2) A change in price can cause a change in the quantity of a product that its users consume. Usually, however, it works both ways. What is the Effect on Cigarette Sales? The reason sales continue to increase in spite of price increases is that price is but one factor in the demand for cigarettes, or anything else, for that matter. One thing that has caused cigarette sales to increase is that there has been an increase in the smoking-age population. Between 1970 and 1980 the population aged 18 to 64 increased at an average annual rate of 1.83 percent. This increase is slowing, however, and will average 0.97 percent between 1980 and 1990. Another thing is that there appears to be a strong secular trend. Nontheless, price is also an important factor influencing cigarette sales, as will be shown below. In attempting to determine which variables are important in determining ciga- rette sales, I went back to my old sales forecasting model and ran repeated step-wise multiple regressions, using various combinations of all of the variables that I could think of that could be remotely related to cigarette sales.3 Three variables proved to be most highly correlated with sales, both singly and in combination. They were the secular trend, population, and price per pack, and the addition of other variables decreased the significance of the correlation. Chart 2 shows the relationship between actual sales and (1) The secular trend alone, (2) The secular trend and population, and (3) The trend, population, and price per pack. As can be seen, the addition of the population variable improved the fit considerably over the simple trend variable, and the addition of price further improves the correspondence between actual and calculated sales. The three variables explain 99.8 percent of the variance (p<.00001). When taken together the significance levels of the three variables are: population p<.005, trend p<.002, and price p<.00005. Thus; price clearly is important. 3. Many of the variables were so highly intercorrelated that factor analysis could not be used to reduce the variables to a few simple factors, hence this more cumbersome method.

3 The effect of a price increase is to cause sales to increase more slowly than would have been the case without a price increase. This means that if the elasticity is -0.42, a ten percent increase in price would cause sales to be 4.2 percent lower than would have been the case if there had not been a price increase. In a period of relative price stability, a price increase would be followed by an absolute decline in sales, but in a period of sustained inflation, when price increases of one good are masked by the general increase in prices, the effect of a price increase should, theoretically, be relatively transitory. This should be particularly true of products (like cigarettes) the prices of which are increasing more slowly than the consumer price index for all commodities. (Curiously, neither the consumer price index nor the ratio between cigarette price increases and overall price increases proved significant when entered into the sales forecasting equation.) A State-by-State Analysis Another way of looking at the effect on sales of price changes is to observe what happens to sales in individual states when the excise taxes are changed. Obviously, not all states can be studied in this manner: The only states that can be used are those in which there are no major population concentrations near the borders, or contiguous states in which the cigarette price differentials have remained about the same over the period of time being studied. States that are heavily dependent on tourism (such as Florida and Nevada) must also be excluded, because of fluctuations in tourist trade. A good illustration of why we must impose this restriction, and also an illustration of a form of cross elasticity, is shown in Chart 3, which shows cigarette prices and per capita sales in Massachusetts and New Hampshire from 1961 to 1981. Massachusetts has a sufficiently large population that price differentials between the two states would be expected to have little impact on per capita cigarette sales there, while New Hampshire, being much smaller, would show greater sensitivity in per capita sales. In the early 1960's cigarettes were about 15 percent more expensive in Massachusetts than in New Hampshire, enough of a difference to make it worthwhile for people in the Boston area to drive to New Hampshire to buy cigarettes. In 1965 both states raised taxes and Massachusetts followed with another increase in 1966, and in the two following years per capita sales in New Hampshire increased sharply. A 1967 increase by New Hampshire returned the relative prices in the two states to about what it was in the 1960-65 period, and over a two year period per capita sales in New Hampshire dropped to about the 1960-65 level. A 1969 increase by Massachusetts caused per capita sales in New Hampshire to begin to rise, and this continued even after a small 1970 increase in taxes in New Hampshire. In 1971 both states increased taxes but the Massachusetts increase was larger, so the upward trend in per capita sales in New Hampshire continued until the gasoline shortage in the winter of 1973-74 made driving to Massachusetts to buy cigarettes less attractive. In July of 1975 both states raised their taxes, New Hampshire by one cent and Massachusetts by five. Again per capita sales in New Hampshire rose, but after that the price differential remained constant and gasoline prices continued to rise, and per capita sales in New Hampshire went into a decline.

4 The remaining charts show the effect tax increases have had on per capita sales in several of the states selected for analysis. (The use of per capita sales instead of total sales corrects for the different rates of population growth in the various states.) Per capita sales are for fiscal years. In California, the 1967 tax increases caused an 8.5 percent drop in sales, and per capita sales never regained their former level. In Minnesota, the 1969 price increase depressed sales for one year, after which they rebounded to the earlier level, and the 1971 increase caused a sharper drop in per capita sales and resulted in a slower increase to previous levels. Colorado is an interesting case, because at one point it decreased taxes, thus making it possible to compute the price elasticity of cigarettes under conditions of price decreases as well as price increases. Both the 1973 and 1977 tax increases caused declines in sales that yield price elasticities of -.22, and the tax decrease in 1978 resulted in increased sales, and again show a -0.22 price elasticity. Many states showed declining price elasticities over time: In New Mexico, the 1968 price increase yields an implied elasticity of -0.56, but the 1961 price increase (not shown) yielded an elasticity of -0.83. In Texas, the elasticity in 1969 was -0.25, and the price increase in 1971 only slightly slowed the underlying upward trend. In contrast, the elasticities were -0.39 in 1959 and -0.68 in 1955. In South Dakota the elasticity in 1969 was -0.51 and in 1979 it was -0.26, but in 1963 it was -0.87. This pattern of declining elasticities over time is apparent in other states as well. In Wisconsin, for example, the implied elasticities in years of price increases were as follows: 1961 -0.63 1963 -0.55 1965 -0.37 1969 -0.31 1971 nil In Oklahoma the elasticity was -0.57 in 1961 and -0.21 in 1968. In Nebraska it was -0.96 in 1963 and -0.20 in 1971. In Utah: -0.74 in 1963 and -0.11 in 1979. In Idaho: -0.80 in 1961 and -0.44 in 1972. Conclusion As noted above, price is but one factor influencing cigarette sales. On the basis of the evidence of this state-by-state analysis, it appears that price elasticities are declining over time, and that the price elasticity of cigarettes now is about -0.2. Thus a ten percent increase in price would cause sales to be about two percent lower than they would have been without a price increase. Inasmuch as the increase in profits occasioned by an industry price r

increase would more than offset any loss due to a slower increase in unit sales, there is clearly no reason to lose any sleep when we raise prices. We are all, however, entitled to our nights of insomnia if the federal government decides to increase the excise tax by any significant amount. You may recall that I reported last fall that a member of the staff of the House Budget Committee told me that the increase being considered was 25 percent, or two cents a pack. A two cent price increase would, theoretically, depress sales by six tenths of one percent. In practice, it would probably go unnoticed, except for all the publicity that would surround it in the congressional debates. If there is a proposed increase, and if it is indeed one dollar per thousand, it would probably be less harmful to the industry to accept it without comment than to fight it and give the anti-smoking fanatics another forum in which to vent their spleen. MEJ : f cc: M. Hausermann L. R. J. T. A. D. C. Meyer Thomson Zoler Goodale Udow Nelson Rowe

6 4 2 0 Chart 1-- Number of States Imposing or Increasing Excise Taxes, by Year, 1950-1981 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 Year gtea9getQz

Chart 2 ® Actual Sales and Calculated Sales Based on Various Varsables 640 620 ~ B i 1 600 J 1 580 i 0 n S 560 540 J 520 J -'-~Trend Only 0 _-r9Trend and Popu l at i on °---° Trend, Pr i ce, and Popu 1 at i on *`--*Actua l Sa 1 es ,.- 70 71 72 73 74 75 76 77 76 79 80 Year & TC9?S£v0z

Chart 3 - Per Capita 5ales (Packs X 10) and Price, New Hampshsre and Massachusetts 60 40 20 0 ~-~ N.H. Price °---° N. H. Sa 1 es °--~Mass. Price *---*Mass. 5ales ~--~-e~---o---~ ~---z^-~-~ --~--," n--e- "B- - ~_ " _© -~.-fl- -.°- _ .®--e- -~ 61 62 63 64 65 66 67 6B 69 70 71 72 73 74 75 76 77 76 79 BO B 1 Year ozegggeaaz

CALIFORNIA (Tax Increased 8/1/67, 10/1/67) 60 J 40 J A Price per Pack 13---OPer Capita Consumpt i on (100s) --e--~p-~-p--- B-, - B- - -fl- -- f1---'a ---8 ---}3-,_ ©--,8-,- Q 20 J 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 Year ~~CS;9QCt07,

MINNESOTA (Tax Increased 5/30/69. 10/31/71) 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 Year